Enigmata: Scaling Logical Reasoning In LLMs With Synthetic Verifiable Puzzles arxiv.org 1 points by optimalsolver 5 hours ago